A computing data center may include one or more computing systems, each including a plurality of compute nodes that may comprise various compute structures (e.g., servers and/or storage systems) and may be physically located in multiple sleds or boards disposed on a computing shelf or computing tray of a rack. The compute nodes may be interconnected, typically via one or more switches, e.g., Network or Ethernet switches, such as a Top-of-Rack (ToR) switch, forming different compute structures.
Typically, network data traffic from compute nodes disposed in a sled may be routed by providing a separate connection for each sled in the computing tray. The sled level data traffic may then be routed to an external computing component, e.g., via a ToR switch, with discrete, fixed-bandwidth cabling. To provide desired connection efficiency, an additional local network switch (e.g., Ethernet switch) may be required. However, this solution may cause difficulties with dynamically accessing, servicing, and interconnecting different network components (e.g., compute nodes) residing in sleds of a computing tray. Further, the current solution may provide for fixed bandwidth distribution among the compute nodes in the sled, which may result in inefficient use of available compute resources.